Google cuts its quantum processor's errors by 40 times

Google Quantum AI published a study in Science on October 15, 2026, titled “Magic state cultivation on a superconducting quantum processor”, describing an experiment conducted on its Willow quantum processor.

The technique, called magic state cultivation, produced what the researchers define as T states with a logical fidelity of 99.99%, reducing errors by about 40 times compared to states produced without this procedure. This represents a step toward so-called fault-tolerant quantum computers, meaning systems capable of continuing to function correctly even in the presence of errors in individual components, one of the main theoretical and practical obstacles to the development of this technology.

Google claims that producing high-purity magic states, without the technique described in the study, previously risked consuming up to 95% of a quantum computer’s overall resources. Reducing this cost is one of the central issues in making quantum systems usable for real-world problems, beyond the small-scale demonstration experiments conducted so far in research laboratories.

It should be noted that the study comes from the same Google Quantum AI researchers who developed both the technique and the Willow processor on which it was tested: the news, although published in a peer-reviewed journal, currently stems from a single source in the material collected, without independent verification by other quantum computing laboratories. This does not invalidate the result, but for now limits its confirmation to a single party, which is also the main beneficiary of the news.

The result is part of a path of progressive improvement in Google’s quantum hardware, with Willow already at the center of previous company announcements on reducing errors in superconducting processors. It remains to be seen whether the technique will be reproduced on architectures other than those developed internally by Google.